ELECTRICAL CONTACTS TO EVAPORATED CADMIUM SELENIDE

The voltage-current characteristics of evaporated gold-CdSe contacts have been experimentally investigated. In contrast to the experiments of Weimer with Cadmium Sulphide, it was found that CdSe makes no ohmic contact with the underlying gold layers. The reverse biased contact has a zener-like behaviour. With silver ink contacts deposited upon the CdSe, perfectly ohmic contacts were obtained.

Firstly the well known theory of a metal n-type semiconductor contact is summarised. The metal and the semiconductor are characterised by their work functions CM and s respectively. For a given semiconductor the energy difference between the zero energy level Eo outside the semiconductor, and the bottom of the conduction band at the surface is a fixed quantity, the electron affinity . When the doping concentration is known, the energy difference Ecb-EF between the bottom of the conduction band and the Fermi level in the bulk can be calculated. The work function s is then: $s X + Ecb E F Initially the case CM;>S is considered. Figure la shows the situation before and after the contact was made. The metal will be negatively charged throughout a shallow surface layer. In the semiconductor a depletion layer is built up. So a diode behaviour of the junction is to be expected. In the other case tM .<bS (fig. b) an accumulation layer will be formed in the semiconductor. This junction shows a non rectifying contact. One says that the junction is ohmic. However, in the case that qM>S an ohmic junction can be formed when the carrier concentration is very high. In this case the thickness of the depletion layer is so small that conduction takes place by quantum mechanical tunneling through the space charge layer Weimer 2,3 has found that gold evaporated on CdS gives no ohmic contact, while CdS evaporated upon gold makes a good ohmic contact. His explanation is, that the first condensed semiconductor layer was Cd rich, and thus heavily doped with n-type charge carriers. The space charge layer becomes very 39 thin and conduction takes place by quantum mechanical tunneling.
In the experiment CdSe was used as semiconducting material and gold for the metallic contacts. The materials were evaporated from a heated tungsten boat in a vacuum of about 10 -6 Torr.A 1" square corning glass was used as substrate. Before the deposition of the semiconductor, gold contacts were evaporated on the cold substrate. The geometry used was a cross shaped Hall generator ( Figure 2) with h 8 mm and 4 mm4. The thicknesses of the semiconductor and the gold layer were 1000 A and 500 A respectively, and the area of the overlapping region was 2 mm 2 in order to assure good contact. The geometry offers the advantage that Hall mobility and carrier concentration on the one hand and I-V characteristics on the other hand can be measured on the same semiconductor sample. If the hypothesis of Weimer were also true for CdSe, the contacts should be ohmic. However the contrary is found. Figure 3 shows the I-V characteristics measured between two opposite contacts for various current ranges. The current was supplied by the Fluke programmable current calibrator, model 3330B. This current source provides a stable current even in the nanoampere range. The potential between the contacts was measured with the Fluke high impedance voltmeter/ null detector model 845A. Because the internal impedance of the voltmeter is 10 M2, some numerical corrections were necessary to find the true current through the sample. For low voltages the sample looks like a resistor of 11 M2. For high values of the voltage the curve has a slope of 113 kf2. When the current goes from the semiconductor towards the metal, this contact is a reverse biased metal-  figure 4. This characteristic is very similar to the voltage-  Other evidence that the assumption may be correct is that the blocking contact was damaged when a current of mA was supplied. This is easily explained by the fact that a heat dissipation of 5 mW then exist in the shallow space charge layer. It is also assumed that the CdSe-metal system may be analysed independently of surface states as has been done with CdS-metal systems a.
One concludes that the conditions in a Au-CdS contact are not fulfilled in a Au-CdSe contact; either the first condensed CdSe layers are not sufficiently Cd-rich, or the difference in work functions is too large.
After these experiments, silver conducting ink was used as the contact material. These contacts were perfectly ohmic as can be seen from Figure 5. Even in the nanoamp6re range the slope of the curve did not vary appreciably. The new value of 230 k2 for the semiconductor resistance was caused by a thermal treatment of the sample (baking at 100C for one hour). Good ohmic contacts were also obtained by using evaporated Aluminium.
The strong dependence of the I-V characteristics on the used metals (Au, A1 or silver ink) proves that the surface states do not play an important role, otherwise the junction characteristics would be less dependent on the metal used, as reported by Pfeifer